Computational Mathematics and Modeling

, Volume 29, Issue 3, pp 359–366

# Analysis of Dissimilarity Set Between Time Series

• A. V. Goncharov
• V. V. Strijov
Article

This paper investigates the metric time series classification problem. Distance functions between time series are constructed using the dynamic time warping method. This method aligns two time series and builds a dissimilarity set. The vector-function of distance between the time series is a set of statistics. It describes the distribution of the dissimilarity set. The object feature description in the classification problem is the set of selected statistics values of the dissimilarity set. It is built between the object and all the reference objects. The additional information about the dissimilarity distribution improves the classification quality. We propose a classification method and demonstrate its result on the classification problem of the human physical activity time series from the mobile phone accelerometer.

## Keywords

time series metric classification dynamic time warping distance function

## References

1. 1.
A. V. Goncharov, M. S. Popova, and V. V. Strijov, “Metric time series classification using dynamic warping relative to centroids of classes,” Systems Means Inform., 25, No. 4, 52–64 (2015).Google Scholar
2. 2.
E. J. Keogh and C. A. Ratanamahatana, “Exact indexing of dynamic time warping,” Knowl. Inform. Syst., 7, No 3, 358–386 (2005).
3. 3.
F. Petitjean, G. Forestier, G. I. Webb, A. E. Nicholson, Y. Chen, and E. Keogh, “Dynamic time warping averaging of time series allows faster and more accurate classification,” in: IEEE Int. Conf. Data Eng. (ICDE), IEEE Computer Society, Chicago (2014), pp. 470–479.Google Scholar
4. 4.
D. J. Berndt and J. Clifford, “Using dynamic time warping to find patterns in time series,” in: Workshop on Knowledge Discovery in Databases, 12th International Conference on Artificial Intelligence, Seattle (1994), pp. 359–370.Google Scholar
5. 5.
E. Frentzos, K. Gratsias, and Y. Theodoridis, “Index-based most similar trajectory search,” in: IEEE International Conference on Data Engineering (ICDE), IEEE Computer Society, Istanbul (2007), pp. 816–825.Google Scholar
6. 6.
M. D. Morse and J. M. Patel, “An efficient and accurate method for evaluating time series similarity,” in: ACM International Conference on Management of Data (SIGMOD), ACM, Beijing (2007), pp. 569–580.Google Scholar
7. 7.
Y. Chen, M. A. Nascimento, B. C. Ooi, and A. K. H. Tung, “SpADe: On shape-based pattern detection in streaming time series,” in: IEEE International Conference on Data Engineering (ICDE), IEEE Computer Society, Istanbul (2007), pp. 786–795.Google Scholar
8. 8.
S. Salvador and P. Chan, “Fastdtw: Toward accurate dynamic time warping in linear time and space,” Workshop on Mining Temporal and Sequential Data, Seattle, 70–80 (2004).Google Scholar
9. 9.
P.-F. Marteau and S. Gibet, “On recursive edit distance kernels with application to time series classification,” IEEE Trans. Neural Netw. Learn. Syst., 1–14 (2014).Google Scholar
10. 10.
D. Haussler, “Convolution kernels on discrete structures,” in: Technical Report UCS-CRL-99-10, University of California at Santa Cruz, Santa Cruz (1999).Google Scholar
11. 11.
B. Scholkopf, A. Smola, and K.-R. Muller, “Nonlinear component analysis as a kernel eigenvalue problem,” Neural Comput., 10, No 5, 1299–1319 (1998).
12. 12.
M. Cuturi, J.-P. Vert, O. Birkenes, and T. Matsui, “A kernel for time series based on global alignments,” in: In Acoustics, Speech and Signal Processing, ICASSP 2007, IEEE International Conference, 2 (2007), pp. 413–416.Google Scholar
13. 13.
Data from accelerometer. Available at: http://sourceforge.net/p/mlalgorithms/TSLearning/data/preprocessed large.csv (accessed November 15, 2016).